Neuroscience Letters 566 (2014) 21–26
Contents lists available at ScienceDirect
Neuroscience Letters journal homepage: www.elsevier.com/locate/neulet
Using ictal high-frequency oscillations (80–500 Hz) to localize seizure onset zones in childhood absence epilepsy: A MEG study Ailiang Miao a , Jing Xiang c , Lu Tang a , Huaiting Ge a , Hongxing Liu a , Ting Wu b , Qiqi Chen b , Zheng Hu d , Xiaopeng Lu d , Xiaoshan Wang a,∗ a
Department of Neurology, Nanjing Brain Hospital, Nanjing Medical University, Nanjing, Jiangsu 210029, China MEG Center, Nanjing Brain Hospital, Nanjing, Jiangsu 210029, China c MEG Center, Division of Neurology, Cincinnati Children’s Hospital Medical Center, 3333 Burnet Avenue, Cincinnati, OH 45220, USA d Department of Neurology, Nanjing Children’s Hospital, Nanjing, Jiangsu 210029, China b
h i g h l i g h t s • • • •
Ictal high-frequency oscillations correlate with absence seizures. The rate of fast ripples is associated with absence seizure frequency. Ictal high-frequency oscillations are mainly located to medial prefrontal cortex. Compared with spikes, high-frequency oscillations have more focal distribution.
a r t i c l e
i n f o
Article history: Received 6 December 2013 Received in revised form 5 February 2014 Accepted 15 February 2014 Keywords: Childhood absence epilepsy High-frequency oscillations Morlet continuous wavelet transform Dynamic magnetic source imaging Magnetoencephalography
a b s t r a c t This study aimed to use ictal high-frequency oscillations (HFOs) ranging from 80 Hz to 500 Hz to locate seizure onset zones in childhood absence epilepsy (CAE) using non-invasive magnetoencephalography (MEG). Ten drug-naïve children with CAE were studied using a 275-channel MEG system. MEG data were digitized at a sampling rate of 6000 Hz. HFO spectral power in real-time spectrograms was assessed using Morlet continuous wavelet transform. Magnetic sources were volumetrically localized through dynamic magnetic source imaging with a slide window. HFOs were identiﬁed in all patients. The total time of fast ripples (250–500 Hz) was greater than that of ripples (80–250 Hz) during absence seizures. The rate of fast ripples was associated with seizure frequency. HFO duration was signiﬁcantly longer when co-occurring with spikes than when occurring independently, and the maximum frequency of HFOs co-occurring with spikes was higher than that of HFOs occurring independently. HFOs were predominantly localized in the medial prefrontal cortex (MPFC), whereas spikes were widespread to a variety of regions during the absence seizures. Compared with spikes, HFOs appeared to be more focal. The ﬁndings indicate that HFOs in the MPFC have a primary function in initializing epileptic activity in CAE. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Childhood absence epilepsy (CAE) is a non-convulsive event characterized by brief (approximately 5–30 s) disruptions in consciousness, as well as 3 Hz bilateral, synchronous spike-and-wave discharges (SWDs) on normal background activity on electroencephalography (EEG) . Although the argument of initialization
∗ Corresponding author at: Department of Neurology, Nanjing Brain Hospital, Nanjing Medical University, Guang Zhou Road 264, Nanjing, Jiangsu 210029, China. Tel.: +86 25 8229 6208; fax: +86 25 8371 9457. E-mail addresses: [email protected]
(A. Miao), [email protected]
, [email protected]
(X. Wang). http://dx.doi.org/10.1016/j.neulet.2014.02.038 0304-3940/© 2014 Elsevier Ireland Ltd. All rights reserved.
in CAE has lasted for more than a decade, the exact absence seizure onset zones (SOZ) have not been fully elucidated. Magnetoencephalography (MEG), a relatively new clinical neuroimaging modality, is well suited for the study of epileptic discharges because MEG can noninvasively detect and localize neuromagnetic signals [2–5]. Brain activity in a low-frequency range ( 0.05). (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of the article.)
domain. If signals appeared in the given sensitive time (a small sigma value) and sensitive frequency (a large sigma value) ranges, they would be enhanced. With a ﬁxed sigma value (number of oscillation), the feature of wavelet is sensitive to the frequency in low frequency while sensitive to time in high frequency. To overcome this weakness and focus on the frequency characteristics of spontaneous brain activation, we improved our time–frequency analysis method by dynamically changing the sigma value (the number of wave circles) according to frequency ranges. To quantify neuromagnetic signals in a wide frequency range with balanced temporal and frequency resolutions, the sigma values considered were 1 and 6 for frequency ranges of 14–70 Hz and 80–500 Hz, respectively. With 600 frequency bins, the frequency resolution was 6.315, 6.318 data points per Hz for the 2 frequency ranges, respectively. For each seizure, CWT was performed in 400 ms time windows over the course of 400 ms before the visualized ictal onset point and ictal period after the onset point. The CWT was applied to each bandwidth: 14–70 Hz and 80–500 Hz separately (see Fig. 1).
C −1 B B T C −1 B
where B is the forward solution for a unit current dipole; indicates a location (a node in grid beamformer) and C is the covariance matrix of time–frequency data. Multiple local spheres were used for magnetic forward computing. One methodological improvement in the present study was that the dynamic magnetic source imaging (dMSI) with a sliding window was used to capture the dynamic spatiotemporal activity over time at ictal recordings rather than at a single time point (e.g., spike peak) with a source localization algorithm in the brain. In prior studies, the dipole modeling and SAM can only localize a single discrete source. The dMSI can scan the entire brain and determine the number of sources automatically, which can clearly show propagative pattern of magnetic sources. Dynamic and volumetric source images were created and integrated into the 3D head model reconstructed brain images using the patients’ own coregistered magnetic resonance imaging (MRI) using a customer-designed program, MEG Processor [3,14].
2.6. Source analysis 3. Results Similar to our previous reports [2–5], wavelet-based beamformer was used to locate neuromagnetic signals. Wavelet-based beamformer includes two main mathematical steps. The ﬁrst step was to transfer MEG waveforms to time–frequency representation with aforementioned Morlet wavelet algorithm (Eq. (1)). The second step was to scan source with a beamformer. At each coordinate voxel in source imaging, the beamformer method computed coefﬁcients W using the following equation:
None of the 10 children recruited in the study had taken any medicine. Therefore, all subjects were drug naïve and the effect of drugs on MEG data could be eliminated. The clinical details are shown in Table 1. The children are aged 5–11 years, with an average age of 8 years. The seizure frequency ranged from 3 to 15 times per day. We successfully obtained 33 stereotyped ictal MEG recordings from the 10 children.
A. Miao et al. / Neuroscience Letters 566 (2014) 21–26
According to our routine MEG system and environmental noise test, the polarity spectrograms computed from MEG data without subjects revealed activities of approximately 50 Hz. Therefore, we considered the neuromagnetic signals at 50 Hz as power-line noise (Fig. 1A). Therefore, they were not analyzed. 3.1. Characteristic of ictal HFOs Spectrograms showed HFOs (80–500 Hz) in all subjects, which suggested the association of absence seizures with HFOs. Spectrograms also revealed that spectral power in 14–70 Hz was signiﬁcantly increased, which probably reﬂected the spikes of the spike–wave discharges of the absence seizure. HFOs and spikes could occur at the same time (co-occurring) or at different time
(independent) (Fig. 1C–F from patient 1). The ictal MEG recording data exhibited that 11% of the time was occupied by ripples and 27% by FRs (Table 1). In absence seizures, FRs appeared more frequently than ripples, and the total FR time was longer than the ripple time (Fig. 1G and H; Table 1). Fig. 1G and H shows that seizure frequency ﬂuctuates according to FRs. Moreover, the seizure frequency increases with the increase in FRs, suggesting that FRs may be associated with seizure frequency (Table 1). The duration of HFOs was signiﬁcantly longer when co-occurring with spikes than when occurring independently, and the maximum frequency of HFOs co-occurring with spikes was higher than for that occurring independently (Fig. 1C–F). A small proportion of HFO-covered ripples and FRs had a frequency of approximately 250 Hz (Fig. 1D). Some HFOs co-occurring with spikes remained
Fig. 2. In (A), HFOs appear locally to the left MPFC, whereas the cortical spike sources in the left MPFC (a) propagate toward the left lateral prefrontal cortex (e) through the thalamus (b–d). Compared with spikes in medial prefrontal cortex, HFOs have more restricted distribution. (B) exhibits HFOs from patients 2–4 have more restricted distribution than spikes. (C) presents the regions generating HFOs from patients 5–8. In the magnetic source imaging (MSI), red and yellow regions indicate the discharging epileptic source, whereas black arrows display the direction of propagation. A indicates anterior; P indicates posterior. (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of the article.)
A. Miao et al. / Neuroscience Letters 566 (2014) 21–26
identiﬁable even after the spikes disappeared (Fig. 1F, yellow and white arrows). In Fig. 1I and J, Pearson correlation analysis displays a positive correlation between the ratio of FRs (r = 0.9533; P < 0.0001) and seizure frequency, but not for ripples (r = 0.4028; P > 0.05). This observation indicates that FR density over time can predict the clinical severity (number) of absence seizures. 3.2. Regions generating ictal HFOs Ictal HFOs were highly localized in the medial prefrontal cortex (MPFC, 10/10) (Fig. 2; Table 1), whereas ictal spikes localized within a wide region (Fig. 2A). Moreover, an obvious spreading tendency of spiking sources was observed, except for HFOs. This ﬁnding suggests that only a small region produced HFOs, and extensive areas of the brain generated spikes during the seizures. However, the highest power magnetic sources in spikes were also predominantly located in the MPFC. In addition, HFOs were located in the lateral prefrontal cortex (2/10) and in the precuneus (2/10). 4. Discussion In this study, HFOs were identiﬁed with polarity spectrogram, and the sources for HFOs were localized to the MPFC with dMSI. The results indicate that HFOs in the MPFC may have a primary function in the absence seizures. 4.1. Relationship between HFOs and absence seizures FRs appeared more frequently than ripples. The total FR time was longer than that of ripples in the ictal period, which suggested that FRs may have a more important function in absence seizures than ripples. FRs appeared to be pathological, except for those in the sensory-evoked potentials [17–19]. FRs are probably generated by “out of phase” neural activity or a group of neurons ﬁring asynchronously because of epileptogenic lesions . In absence epilepsy, FRs may also indicate an ictal electrophysiological correlate to subtle motor signs during absence seizures. Such motor signs are the craniuo-caudal march in Snead III “Absence Seizures” from “Epilepsy: A Comprehensive Textbook” . Urrestarazu et al.  reported a clear relation between spikes and FRs, namely, FRs were more restricted to the presumed SOZ than ripples. A small proportion of HFOs covered the ripple and FRs, and the frequency was approximately 250 Hz (Fig. 1D). Furthermore, differentiation by frequency band was insufﬁcient to distinguish between ripples and FRs. In this study, we observed that the rate of FRs was associated with absence seizure frequency. In rats, the rate of HFOs correlated with the rate of seizures , suggesting that HFOs not only identify the SOZ but also reﬂect seizure propensity. Moreover, the rate of HFOs increases in a manner similar to that of seizures when the dosage of anticonvulsive drugs was reduced in patients with epilepsy . In this study, we observed that the duration of HFOs was significantly longer when co-occurring with spikes, and the frequency was higher than those that occurred independently. Urrestarazu et al.  also reported HFOs primarily co-occurred with epileptic spikes, and the power at high frequencies was observed to be reduced after spikes. Differences in duration between HFOs at different time points may result from different strengths of gammaaminobutyric acid (GABA)-mediated inhibition . This hypothesis was strengthened by recent evidence from Jones and Barth , which showed that neocortical FR duration increased after the application of the GABAA receptor antagonist (bicuculline) during epileptogenesis. Therefore, HFOs and spikes generated in the same neural network may increase both the occurrence likelihood and the duration of HFOs.
4.2. Identiﬁcation of SOZ The precise locations of the initialization of SWDs have been debated on for more than a decade. The unifying hypothesis that the SWDs of absence seizures are probably produced through reciprocally connected neurons in the thalamus and cortex is currently generally accepted. However, whether the cortex or thalamus has the primary function remains unknown. HFOs have been extensively investigated as new markers of epileptogenicity. In this study, dMSI showed that HFOs were highly localized in the MPFC and that spikes were more diffusively localized during the ictal period. A previous study that used intracranial depth stereo-EEG investigations reported that HFOs remained conﬁned in the same area (possibly epileptogenic area) during interictal and ictal periods, whereas spikes were more widespread during seizures than interictal period . HFO-generating regions identiﬁed SOZ with greater speciﬁcity and accuracy compared with that of the spike-generating regions [9,27]. Furthermore, some studies indicated that removal of the HFO-generating region predicted better surgical outcome [10–13]. These studies suggested that HFOs appeared to be reliable biomarkers of SOZ. Therefore, the ﬁndings of the present study reliably show that the HFOs are localized in MPFC, and MPFC has a crucial function in the initialization. Furthermore, studies in humans with absence spells have established the importance of MPFC in absence discharges [28–31]. Our results support the notion that the cortex has a primary function in the initialization of absence seizures, consistent with some previous reports [28–31]. We conclude that CAE is associated with HFOs (80–500 Hz). FRs are more associated with absence seizures than ripples. In addition, the rate of FRs is associated with seizure frequency. MPFC has a primary function in the initialization of absence seizure. HFOs have a more restricted distribution compared with spikes. Acknowledgements This study was supported by Key Project of Medical Science and Technology Development Foundation, Nanjing Department of Health (No. ZKX11002, http://www.njh.gov.cn/html/list 83.shtml). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. References  Proposal for revised classiﬁcation of epilepsies and epileptic syndromes. Commission on Classiﬁcation and Terminology of the International League Against Epilepsy, Epilepsia 30 (1989) 389–399.  R. Kotecha, M. Pardos, Y. Wang, T. Wu, P. Horn, D. Brown, D. Rose, T. DeGrauw, J. Xiang, Modeling the developmental patterns of auditory evoked magnetic ﬁelds in children, PLoS ONE 4 (2009) e4811.  R. Kotecha, J. Xiang, Y. Wang, X. Huo, N. Hemasilpin, H. Fujiwara, D. Rose, T. DeGrauw, Time, frequency and volumetric differences of high-frequency neuromagnetic oscillation between left and right somatosensory cortices, Int. J. Psychophysiol. 72 (2009) 102–110.  J. Xiang, Y. Liu, Y. Wang, E.G. Kirtman, R. Kotecha, Y. Chen, X. Huo, H. Fujiwara, N. Hemasilpin, K. Lee, F.T. Mangano, J. Leach, B. Jones, T. DeGrauw, D. Rose, Frequency and spatial characteristics of high-frequency neuromagnetic signals in childhood epilepsy, Epileptic Disord. 11 (2009) 113–125.  J. Xiang, Y. Wang, Y. Chen, Y. Liu, R. Kotecha, X. Huo, D.F. Rose, H. Fujiwara, N. Hemasilpin, K. Lee, F.T. Mangano, B. Jones, T. DeGrauw, Noninvasive localization of epileptogenic zones with ictal high-frequency neuromagnetic signals, J. Neurosurg. Pediatr. 5 (2010) 113–122.  A. Bragin, J.J. Engel, C.L. Wilson, I. Fried, G. Buzsaki, High-frequency oscillations in human brain, Hippocampus 9 (1999) 137–142.  R.J. Staba, C.L. Wilson, A. Bragin, I. Fried, J.J. Engel, Quantitative analysis of highfrequency oscillations (80–500 Hz) recorded in human epileptic hippocampus and entorhinal cortex, J. Neurophysiol. 88 (2002) 1743–1752.  E. Urrestarazu, R. Chander, F. Dubeau, J. Gotman, Interictal high-frequency oscillations (100–500 Hz) in the intracerebral EEG of epileptic patients, Brain 130 (2007) 2354–2366.  J. Jacobs, P. LeVan, R. Chander, J. Hall, F. Dubeau, J. Gotman, Interictal highfrequency oscillations (80–500 Hz) are an indicator of seizure onset areas
A. Miao et al. / Neuroscience Letters 566 (2014) 21–26 independent of spikes in the human epileptic brain, Epilepsia 49 (2008) 1893–1907. J. Jacobs, M. Zijlmans, R. Zelmann, C.E. Chatillon, J. Hall, A. Olivier, F. Dubeau, J. Gotman, High-frequency electroencephalographic oscillations correlate with outcome of epilepsy surgery, Ann. Neurol. 67 (2010) 209–220. W. Boling, Y. Aghakhani, F. Andermann, V. Sziklas, A. Olivier, Surgical treatment of independent bitemporal lobe epilepsy deﬁned by invasive recordings, J. Neurol. Neurosurg. Psychiatry 80 (2009) 533–538. J.Y. Wu, R. Sankar, J.T. Lerner, J.H. Matsumoto, H.V. Vinters, G.W. Mathern, Removing interictal fast ripples on electrocorticography linked with seizure freedom in children, Neurology 75 (2010) 1686–1694. H. Fujiwara, H.M. Greiner, K.H. Lee, K.D. Holland-Bouley, J.H. Seo, T. Arthur, F.T. Mangano, J.L. Leach, D.F. Rose, Resection of ictal high-frequency oscillations leads to favorable surgical outcome in pediatric epilepsy, Epilepsia 53 (2012) 1607–1617. X. Huo, J. Xiang, Y. Wang, E.G. Kirtman, R. Kotecha, H. Fujiwara, N. Hemasilpin, D.F. Rose, T. Degrauw, Gamma oscillations in the primary motor cortex studied with MEG, Brain Dev. 32 (2010) 619–624. J. Gotman, V. Levtova, B. Farine, Graphic representation of the EEG during epileptic seizures, Electroencephalogr. Clin. Neurophysiol. 87 (1993) 206–214. X. Huo, Y. Wang, R. Kotecha, E.G. Kirtman, H. Fujiwara, N. Hemasilpin, T. Degrauw, D.F. Rose, J. Xiang, High gamma oscillations of sensorimotor cortex during unilateral movement in the developing brain: a MEG study, Brain Topogr. 23 (2011) 375–384. G. Curio, B.M. Mackert, M. Burghoff, J. Neumann, G. Nolte, M. Scherg, P. Marx, Somatotopic source arrangement of 600 Hz oscillatory magnetic ﬁelds at the human primary somatosensory hand cortex, Neurosci. Lett. 234 (1997) 131–134. G.A. Worrell, A.B. Gardner, S.M. Stead, S. Hu, S. Goerss, G.J. Cascino, F.B. Meyer, R. Marsh, B. Litt, High-frequency oscillations in human temporal lobe: simultaneous microwire and clinical macroelectrode recordings, Brain 131 (2008) 928–937. J.J. Engel, A. Bragin, R. Staba, I. Mody, High-frequency oscillations: what is normal and what is not? Epilepsia 50 (2009) 598–604.
 M. Zijlmans, P. Jiruska, R. Zelmann, F.S. Leijten, J.G. Jefferys, J. Gotman, Highfrequency oscillations as a new biomarker in epilepsy, Ann. Neurol. 71 (2012) 169–178.  J. Engel Jr., T.A. Pedley (Eds.), Epilepsy: A Comprehensive Textbook, LippincottRaven Publ., Philadelphia, 1997, p. 579 (Chapter 51).  A. Bragin, C.L. Wilson, J. Almajano, I. Mody, J.J. Engel, High-frequency oscillations after status epilepticus: epileptogenesis and seizure genesis, Epilepsia 45 (2004) 1017–1023.  M. Zijlmans, J. Jacobs, R. Zelmann, F. Dubeau, J. Gotman, High-frequency oscillations mirror disease activity in patients with epilepsy, Neurology 72 (2009) 979–986.  E. Urrestarazu, J.D. Jirsch, P. LeVan, J. Hall, M. Avoli, F. Dubeau, J. Gotman, Highfrequency intracerebral EEG activity (100–500 Hz) following interictal spikes, Epilepsia 47 (2006) 1465–1476.  M.S. Jones, D.S. Barth, Effects of bicuculline methiodide on fast (>200 Hz) electrical oscillations in rat somatosensory cortex, J. Neurophysiol. 88 (2002) 1016–1025.  M. Zijlmans, J. Jacobs, Y.U. Kahn, R. Zelmann, F. Dubeau, J. Gotman, Ictal and interictal high frequency oscillations in patients with focal epilepsy, Clin. Neurophysiol. 122 (2011) 664–671.  L. Andrade-Valenca, F. Mari, J. Jacobs, M. Zijlmans, A. Olivier, J. Gotman, F. Dubeau, Interictal high frequency oscillations (HFOs) in patients with focal epilepsy and normal MRI, Clin. Neurophysiol. 123 (2012) 100–105.  D.M. Tucker, M. Brown, P. Luu, M.D. Holmes, Discharges in ventromedial frontal cortex during absence spells, Epilepsy Behav. 11 (2007) 546–557.  H. Stefan, A. Paulini-Ruf, R. Hopfengartner, S. Rampp, Network characteristics of idiopathic generalized epilepsies in combined MEG/EEG, Epilepsy Res. 85 (2009) 187–198.  K. Sakurai, Y. Takeda, N. Tanaka, T. Kurita, H. Shiraishi, F. Takeuchi, S. Nakane, K. Sueda, T. Koyama, Generalized spike–wave discharges involve a default mode network in patients with juvenile absence epilepsy: a MEG study, Epilepsy Res. 89 (2010) 176–184.  D. Gupta, P. Ossenblok, G. van Luijtelaar, Space-time network connectivity and cortical activations preceding spike wave discharges in human absence epilepsy: a MEG study, Med. Biol. Eng. Comput. 49 (2011) 555–565.